sgdWt_convexLinComBounded: Perform first order stochastic gradient descent update of...

In benkeser/onlinesl:

Description

This function performs a single step of gradient descent on the weight vector for the super learner weights and projects the resulting vector onto the L1-simplex via the internal function .projToL1Simp. The function returns the updated weight vector.

Usage

sgdWt_convexLinComBounded(Y, slFit.t, p.t, tplus1, stepSize = NULL, lower,
  upper)

Arguments

Y	The outcome at iteration t
slFit.t	A named list with a component named alpha.t that contains the 1-column matrix of current estimate of the super learner weights
p.t	The predictions from the various online algorithms at time t
tplus1	The iteration of the online algorithm
stepSize	The size of the step to take in the direction of the gradient. If stepSize=NULL (default) the function uses 1/tplus1.

Value

alpha A matrix of updated weights.

benkeser/onlinesl documentation built on May 12, 2019, 12:09 p.m.